This document presents a novel motion monitoring system using inertial measurement units to track activities of daily living. It describes the system's quaternion algorithm and SensorTag device for collecting acceleration, gyroscope, and magnetometer data in real-time. Experimental results show the system can accurately track linear displacement but struggles with reconstructing eating gestures. Future work will focus on improving angular and displacement computations for wider smartphone applications.
8. 2. Novel motion monitoring system
Quaternion algorithm
Quaternion
šµ
š“
š = [š1 š2 š3 š4] = cos
š
2
ā šš„ sin
š
2
ā šš¦ sin
š
2
ā šš§ sin
š
2
Gyroscope
Angular rate
Accelerometer
Direction of gravity
Magnetometer
Direction of magnetic field
Defection angles roll (X), pitch (Y) and yaw (Z)
14. 2. Novel motion monitoring system
Sensorsā sampling rates and data range
15. 2. Novel motion monitoring system
SensorTag firmware flowchart
ā¢ Initialize the device and advertise
ā¢ Turn on sensors and send data
ā¢ Disable sensors and back to
advertising status
16. 2. Novel motion monitoring system
SensorTag connection with smartphone and laptop
17. 2. Novel motion monitoring system
BLE connection with smartphone
18. 2. Novel motion monitoring system
LabVIEW program flowchart
ā¢ Establish the connection
ā¢ Read and display signals
ā¢ Save data and disconnect
19. 2. Novel motion monitoring system
Signalsā display and save
20. 1. Activities of daily living (ADLs)
2. Novel motion monitoring system
3. Tests and results
4. Conclusion
5. Future work
Content
35. 4. Conclusion
ā¢The displacement measuring results are good for
linear motions.
ā¢The algorithm failed to revise the direction during
eating gesture reconstruction.
What are ADLs and whatās the purpose of applying a motion monitoring system to healthcare?
Very hard for the elderly, disabled, and chronic disease patients.
A real-time ADLs motion monitoring system can help.
Markerless and marker-based mo-cap system are used for animation and filming. They are set up in camera studios and very expensive, although they are accurate. On the other hand, IMU-based system is cheaper and wearable.
Markerless and marker-based mo-cap usually have nine or more cameras. Depth sensing like Microsoft Kinect only has three cameras at most.
Electromyography (EMG) measures the electrical activity produced by skeletal muscles.
Quaternion is used to represent the orientation including three types of deflection angles roll (X), pitch (Y) and yaw (Z).
Markerless and marker-based mo-cap usually have nine or more cameras. Depth sensing like Microsoft Kinect only has three cameras at most.
Electromyography (EMG) measures the electrical activity produced by skeletal muscles.
The direction angles is computed in quaternion, and then use the accelerometer data for acceleration, velocity and displacement calculation.
Small to be wearable.
Only use three sensors. CC2541 is an ultra low power system on chip (SoC) Bluetooth low energy (BLE) solution.
The sampling rate ranges of accelerometer and gyroscope are very wide, but gyroscope consumes very much power with 200Hz or higher, and the magnetometer can only achieve 80Hz at most, thus I only tests up to 100Hz to collect enough data and gain more power savings.
Operating System Abstraction Layer (OSAL) is used in Bluetooth low energy (BLE) protocol. It schedules events in BLE application, and the eleventh one is SensorTag event to perform read and write.
Classic Bluetooth canāt communicate with BLE or Bluetooth smart. Bluetooth smart ready devices can pair with two previous kinds of devices.
Up to seven slave devices can be connected to one central device.
10 seconds to scan for devices and reading starts in 5 seconds after connection.
Data and timestamp save in txt files.
Test the system with displacement measuring error rates, and try to rebuild the two ADLs.
Motion capture lab with 9 cameras. Move the SensorTag along the ruler for 60cm and stop for 1 second every 10cm.
The hand failed to move the SensorTag in a straight line.
Prove the potential to rebuild the motion
Prove the potential to rebuild the motion
Walk 13ft, which is 3.9624m. The ceramic tile is a 1ft-side-length square.
Wrists moved shorter for relative displacement.
For stable long-term angular calculation, it might help to use a better gyroscope and the algorithm needs more improvements.
Develop a better orientation algorithm based on quaternion.
Increase the three sensorsā sampling rates.
Apply CC2650 with better accelerometer, gyroscope and magnetometer or to build a customized circuit.
Develop a smartphone application to connect six SensorTags.